The Design And Implementation Of Ship's Trajectory Analysis System Based On TRACLUS Algorithm

Automatic Identification System(AIS)plays a very important role in nowadays shipping industry,which is used to maintain maritime traffic safety and manage ship transportation.Ships record and transfer data by AIS equipment,which contains a large amount of ship trajectory information.Different from the traditional using methods of AIS data,implied information can be extracted by the way of data mining and be used for analyzing the ship's route characteristics and revealing the ship's behavior rules,so that can provide the more technical ways and reference for maritime traffic management and route planning.In this paper,based on the research and application of AIS data mining at home and abroad,we decide to adopt a trajectory clustering algorithm based on a Partition-and Group Framework,also called TRACLUS algorithm,as the main method,and deeply explore the advantages,basic theory,implementation process and parameter evaluation of this algorithm.The algorithm contains two main parts: trajectory partitioning and line segment clustering.Then,the main characteristic information of the analysis results is expressed by generating representative trajectory.Compared with other algorithms,TRACLUS algorithm will partitioned a trajectory approximately into a set of line segments according to the MDL principle,and then cluster the sub trajectory segments based on the idea of DBSCAN.So that more common characters can be found from the sub trajectory.In this paper,the algorithm is implemented by Python 3 programming language,and we developed it in order to support the spatial data format of AIS trajectories.In order to analyze the ship's trajectory more conveniently and intuitively using TRACLUS algorithm,we designed and developed a WebGIS prototype system based on B/S architecture.In this paper,the design pattern,development framework and deployment process of the system are introduced,and the analysis results of each stage of the algorithm are visualized in the map.In the end,we tested a set of ship's trajectory data in some sea area of British on the system,and compared and analyzed the influence of different parameter values on the clustering results,then the principle of parameter selection is summarized.The experimental results show that the function of the system and the algorithm meet the needs,the ideal clustering effect can be achieved by reasonable parameters.